Single-ply membrane seam QA: inspecting TPO, PVC, and EPDM roof seams

Single-ply seams, and why the seam is the roof

A single-ply membrane seam is the bonded overlap where two sheets of roofing membrane join into one watertight surface. On a single-ply roof, that seam is the whole ballgame. The field of the sheet is a continuous waterproof material that almost never fails on its own. Water gets in at the joints, the flashings, and the penetrations, which is exactly where the membrane stops being one piece and starts depending on a bond someone made by hand or by machine.

So the seam QC is the roof QC. Walk a finished single-ply roof and the field is the part you can stop worrying about. The lineal feet of seam, the pipe boots, the curb flashings, the inside and outside corners, and the T-joints where three layers meet are the entire risk, and they are where every minute of inspection time should go.

The trade splits into two families that seam in two completely different ways. TPO and PVC are thermoplastics, and their seams are hot-air welded so the two sheets fuse into a single material. EPDM is a thermoset rubber that cannot be welded, so its seams are made with splice tape and primer. The inspection looks different for each, but the principle is the same. Find the weak seam before the rain does, and document that you did.

Where a single-ply roof actually leaks

The leaks cluster, and they cluster off the field of the membrane. The first thing an experienced inspector looks at is the flashings and the terminations, not the open sheet, because that is where the overwhelming majority of single-ply leaks start. A roof can have an immaculate field and still leak like a sieve at a rushed pipe boot.

Rank the risk and you spend your time right. Penetrations come first: pipes, conduit, drains, and anything that pokes through the membrane and has to be sealed by hand. Curbs and walls come next, where the membrane turns up vertical and a termination bar and sealant have to hold. The field seams come third, the long welded or taped runs, with the highest risk at the T-joints where seams cross. The open field of the sheet is dead last, and it rarely makes the list at all.

This is why a seam inspection is not a formality you do at the end. It is the inspection. A clean flood test on the field tells you almost nothing if nobody probed the welds and nobody looked hard at the corners. The water finds the one bad detail, and the one bad detail is never in the middle of the sheet.

How do TPO, PVC, and EPDM seams differ?

TPO and PVC seams are hot-air welded; EPDM seams are taped. That single distinction drives everything about how you inspect them. TPO (thermoplastic polyolefin) and PVC (polyvinyl chloride) are thermoplastics, meaning they soften and re-melt with heat, so a hot-air welder fuses the overlapping sheets into one continuous material. A good weld is not a glued joint. It is the two sheets becoming one.

EPDM (ethylene propylene diene monomer) is a thermoset rubber. It is vulcanized, it does not re-melt with heat, and you cannot weld it. EPDM seams are made with a butyl splice tape and a primer, and historically with liquid adhesives before tape took over. The bond is an adhesive bond between two cured surfaces, not a fusion, so it is only as good as the prep and the pressure that set it.

The inspection follows the chemistry. On TPO and PVC you are checking a weld: width, continuity, the squeeze-out bead, and a destructive sample that should tear the sheet instead of opening the seam. On EPDM you are checking an adhesive splice: that the surfaces were primed, that the tape is fully rolled, and that nothing telegraphs a void or a wrinkle under the lap. Confirm the membrane type and the manufacturer's seam method before you write a word, because the acceptance criteria are not interchangeable.

Hot-air welding: the robot and the hand welder

Hot-air welding fuses thermoplastic seams with a stream of hot air that melts the facing surfaces of the two sheets, followed immediately by a roller that presses them together while the polymer is molten. The work splits between two tools. The robotic welder, an automatic machine that rides the seam on its own wheels, runs the long straight field laps at a set temperature and speed. The hand welder, a handheld gun paired with a silicone hand roller, does everything the robot cannot reach: the corners, the details, the patches, and the tie-ins.

Temperature and speed are a matched pair, and they are set to the membrane, not to a habit. Thermoplastic welding commonly runs in the range of 900 to 1100 degrees F at the nozzle, with an automatic welder moving at roughly 8 to 12 ft per minute depending on membrane thickness and ambient conditions. Those are typical figures. The actual setting comes off the manufacturer's welding guidelines and the morning's trial weld, because air temperature, wind, and humidity all move the dial.

The two settings trade against each other. Hot and slow risks scorching and degrading the polymer; cool and fast risks a weld that never fully melted. The skill is finding the window where the sheets fuse without burning, and then re-checking it every time the weather changes. A robot dialed in at 8 a.m. on a cool calm morning is not dialed in at 2 p.m. in full sun with a breeze.

How wide should a hot-air weld be?

A finished hot-air weld is commonly specified at a minimum of 1.5 in wide. That is the continuous, fully bonded width of fused material, not the overlap of the sheets, which is wider. Many manufacturers require any weld measured narrower than the minimum to be overlaid with a cover strip, even if it probes tight, because a narrow weld has no margin for the next dirty spot or cool pass. Confirm the exact minimum and the overlay rule against the membrane manufacturer's specification.

The squeeze-out bead is the field tell of a good weld. When the heat and roller pressure are right, a thin bead of molten polymer is pushed out along the leading edge of the seam, commonly about 1/8 in of consistent flow. That bead means the surfaces actually reached melt and fused. No squeeze-out means not enough heat or pressure, and the weld is suspect before you even probe it.

Too much squeeze-out is its own warning. A heavy bead, wider than roughly 3/16 in, or any sign of smoking, blackening, or the membrane bleeding out, means the welder is too hot and the polymer is being degraded rather than fused. An over-welded seam can look beefy and still be weaker than a clean one, because the heat has cooked the very polymer that was supposed to carry the bond. Read the bead, both directions: too little and too much both fail.

What is a daily test weld and why pull one?

A daily test weld, or trial weld, is a short sample seam welded on scrap membrane at the start of each day and again at every shift change or condition change, then pulled apart by hand to confirm the welder settings before any production seam gets run. It is the single most useful habit on a thermoplastic roof, and it is the first thing a good inspector asks to see.

The acceptance is film-tearing bond, FTB. You cut a strip across the test seam, commonly about 1 in wide, and peel it. A correct weld does not open at the seam. The membrane tears, delaminates, or stretches before the weld lets go, which means the bond is stronger than the sheet itself. If the seam peels cleanly apart at the interface, the settings are wrong and nothing should be welded in production until they are fixed.

Pull a fresh trial weld whenever anything changes: a new roll, a new welder operator, a temperature swing, the wind picking up, clouds rolling in, or the start of the afternoon. The morning weld does not certify the afternoon. Manufacturers also commonly want trial welds logged at a set interval, and that log is part of the warranty file. The trial weld is cheap. The reweld of a thousand feet of bad seam is not.

How do you probe a single-ply seam?

You probe a seam by drawing a rounded, blunt tool along the leading edge of the weld under firm steady pressure, feeling for any spot where the tip slips into the seam. The standard non-destructive test on a thermoplastic roof is to probe every lineal foot of seam this way. Where the probe catches and enters, there is a void, a skip, or a cold weld, and that spot gets marked and repaired.

The tool matters and so does the technique. Use a rounded screwdriver, a cotter-pin puller, or a purpose-made seam probe with a blunted tip, never a sharp point, because a sharp tool will score and damage a good weld and create the very defect you are hunting. Run it along the edge of the weld, not flat across the face, with enough pressure to find a real void without gouging sound material.

Timing is the rookie trap. Probe only after the weld has fully cooled, commonly at least 20 minutes, because a warm seam is still soft and the probe will tear a perfectly good weld and read it as a failure. Probe the whole roof, not a sample. The inspector's walk on a single-ply roof is a probe in one hand and a marker in the other, every foot of seam, every detail, with the flagged spots circled for the repair crew before anyone calls the seam done.

Destructive and non-destructive seam testing

Seam testing comes in two kinds, and a real QC program uses both. Non-destructive tests check the seam without cutting it, so you can run them across the whole roof. Destructive tests cut a sample out, so you run them at intervals and patch the hole. The probe is the workhorse non-destructive test; the cut-and-peel is the workhorse destructive one.

The destructive sample is the proof. You cut a strip across the finished seam, commonly about 1 in wide, and peel it apart, ideally at a steady rate around 2 in per minute, the rate the ASTM peel methods specify, or per the manufacturer's method, looking for the same film-tearing bond the trial weld has to show. The membrane should fail before the weld. A peel sample that opens at the interface condemns the seams welded under those settings, and the cut gets patched with a target over a clean, welded repair.

Other tests fill in around the edges. The air-lance directs a jet of compressed air along the seam edge to lift any unbonded lap, useful as a fast scan. The vacuum-box test sets a clear chamber with soapy water over a seam, pulls a vacuum, and watches for bubbles that show a leak path, useful at details and terminations. Electronic leak detection, the high-voltage spark test and low-voltage vector mapping, finds breaches through the whole membrane rather than just the lap, and it ties into the coating and integrity work covered in the companion roofing guides. Match the test to the question: probe for the field welds, vacuum-box or ELD for the details and the whole-membrane breaches.

What causes a cold weld?

A cold weld is a seam that looks welded but never reached full fusion, so it has little or no real bond and opens under a probe or under load. It is the most common and most dangerous thermoplastic seam defect, because it hides. The seam looks closed, the roof passes a glance, and the lap lets go a season later when nobody is watching.

The causes are a short list and they are mostly preventable. Welding too cold or too fast is the classic pair, where the nozzle never melted the surfaces or the welder outran the heat. A dirty membrane is next: dust, dirt, grease, or a weathered chalky surface sits between the sheets as a boundary layer and stops them from fusing, which is why aged or soiled membrane has to be cleaned with the manufacturer's cleaner and wiped dry before welding. Moisture is the third: dew, condensation, or a damp surface flashes to steam under the heat and blows the bond.

Weather drives most of it. Welding in cold ambient temperatures, commonly a problem below about 40 degrees F without preheating, makes a consistent weld hard because the sheet is cold and the heat bleeds away. Wind and full sun shift the effective heat at the seam from one hour to the next. The over-welded scorch is the opposite failure, where chasing a cold weld with more heat degrades the polymer instead. The defense against all of it is the same: clean dry membrane, settings matched to the conditions, a fresh trial weld whenever the weather moves, and a probe behind every operator.

EPDM seams: splice tape, primer, and the roller

EPDM seams are made with splice tape and primer, and they fail at the prep, not the tape. A modern EPDM field seam is a butyl-based splice tape sandwiched between the two cured rubber sheets, with the lap area first cleaned and coated with a tape primer that activates the bond. The tape carries the watertightness; the primer makes the cured rubber willing to hold it. Skip or shortcut the primer and the tape never really grabs.

The sequence is exact and each step shows up later if it is missed. Clean the splice area, apply the primer and let it dry to the touch with no transfer, commonly under 20 minutes depending on temperature and humidity, position the tape along the marked edge so a small amount, around 1/8 in, extends past the seam edge, then immediately roll it with a steel or silicone seam roller working perpendicular to the seam edge. The roll is not optional. Hand pressure alone leaves voids; the roller is what wets the tape fully into the rubber and drives out the air.

The failure modes read straight off the missed step. Tape that was not primed peels at the rubber face. Tape that was not rolled traps air and leaves channels under the lap. A wrinkle or fishmouth rolled into the seam is a leak path waiting for water. And the older liquid-adhesive seams on aging EPDM roofs are the ones most likely to be opening up now, since adhesive splices were the system's historic weak point long before tape became standard. On an EPDM repair or recover, treat every old adhesive seam as suspect and probe or lift-test it the way you would a questionable weld.

T-joints and the patch

A T-joint is where one seam crosses another, so three layers of membrane stack up and the edges of the cut sheets meet in the middle. It is the highest-risk point on any field seam, welded or taped, because the step created by the layered edges leaves a tiny channel right at the intersection that a straight seam cannot close on its own. Water finds that step.

The fix is a dedicated T-joint treatment, and both families have one. On TPO and PVC, the corner of the cut sheet is commonly skived or the intersection is covered with a welded target patch so the molten polymer flows over and seals the step. On EPDM, a manufactured T-joint patch, a piece of uncured flashing laminated to butyl tape, is primed and rolled over every spot where splice tapes cross or change angle. The patch is what bridges the three-layer step the running seam leaves open.

On the QC walk, every T-joint gets eyes and a probe, and every required patch gets confirmed present and bonded. The mistake is a crew that welds or tapes the long runs cleanly and treats the intersections as just more of the same. They are not. The T-joint is the spot most likely to leak on an otherwise sound seam, and an inspector who skips them has skipped the part that matters most. Confirm the manufacturer's T-joint detail, because the required treatment differs by system.

Flashings and penetrations: where most leaks start

Flashings and penetrations are where the membrane stops being a flat sheet and has to be fitted around something by hand, and that is where most single-ply leaks begin. The pipe boots, the curbs, the inside and outside corners, the drains, and the wall terminations are all hand details, and hand details are where a tired crew at the end of a day cuts the corner that becomes the callback.

Each detail has its own weak point. A pipe boot, prefabricated or field-wrapped, depends on a clean weld or splice at the base and a tight clamp or sealant band at the top. Corners are the worst of it: a prefabricated molded corner is more reliable than a field-fabricated one, because field-fabricating a watertight inside or outside corner out of flat sheet is genuinely hard and rarely done as well twice. Where the membrane turns up a wall or a curb, a termination bar mechanically fastens the top edge and a bead of sealant or counterflashing caps it, and that bar plus sealant is doing real work, not decoration.

Curbs deserve a second look for drainage, not just for the flashing. Water that dams against the uphill face of a wide curb sits right where the flashing is and tests it constantly, which is why a curb in the flow path wants a cricket to split the water around it, the tapered-insulation work covered in the companion guide on roof crickets. On the seam walk, the flashings get the most inspection time and the most skepticism, because the field of the roof was made by a machine and these were made by a person at the end of a shift.

Mechanically attached, fully adhered, or ballasted?

A single-ply roof is held down one of three ways, and the choice changes what the seam has to do. A mechanically attached system fastens the membrane to the deck with screws and plates, commonly in rows that fall along or under the seam laps. A fully adhered system glues the whole membrane down with adhesive and has no field fasteners. A ballasted system lays the membrane loose and holds it with stone or pavers.

The attachment is not just a wind decision; it sets where the load lands. In a mechanically attached system, the fasteners live in the seam, so the seam carries both the watertightness and the structural attachment in the same lineal feet. In a fully adhered system, the adhesive across the whole sheet shares the load and the seam is asked to do less structurally. Fully adhered systems generally achieve higher wind-uplift ratings for that reason, while mechanically attached systems depend heavily on fastener pattern and density.

For inspection, know which system you are on before you walk it. On a mechanically attached roof, the fastener rows under the seam are part of the seam QC: right pattern, right density, plates seated and not overdriven, and the weld or tape covering them intact. On a fully adhered roof, the adhesive bond and the seam are separate checks. On a ballasted roof, the ballast distribution and the loose-laid seam both matter. Confirm the attachment method and the required fastening pattern against the wind-uplift design and the manufacturer's system, because the pattern is engineered, not chosen on the roof.

Why the seam carries the wind uplift load

In a mechanically attached system, the seam is the structural connection, so it carries the wind uplift load directly. Wind moving over a roof creates suction that tries to lift the membrane off the deck. In a mechanically attached system the only things holding the sheet down are the fasteners, and those fasteners sit in the seam laps, so the uplift force travels through the weld or tape into the plate and the screw. A bad seam is not just a leak risk there. It is a structural failure path.

This is why fastener pattern and seam quality are one inspection, not two, on a mechanically attached roof. The wind design, run under ASCE 7 for the wind loads and commonly against FM Global data sheets on insured roofs, sets the fastener spacing and density, with the pattern tightening at the corners and perimeter where uplift is highest. A weld that probes open over a fastener row is a place where the wind can start to peel the sheet, not just where water can enter.

Fully adhered and ballasted systems spread the uplift differently, the adhesive across the whole sheet or the weight of the ballast, so the seam is asked to do less structurally. The lesson holds regardless: confirm the attachment and the fastening pattern against the engineered wind-uplift requirement, ASCE 7 and the applicable FM data sheet or SPRI wind standard, and treat the seam over a fastener row on a mechanically attached roof as load-bearing, because it is.

The QC walk and the punch list

The QC walk is a systematic pass over every seam and every detail with a probe, a marker, and a list, done before the roof is called complete. It is not a stroll. The inspector walks each field seam edge with the probe, marks every catch, then works the details, the pipe boots, the curbs, the corners, the drains, and the terminations, lifting and checking each one. Every flagged spot goes on the punch list with a location.

The repairs come back as welds and patches, not caulk. A void or cold weld on a thermoplastic seam gets rewelded where it can be reached or covered with a welded target patch, then reprobed. An open EPDM splice gets cleaned, primed, and re-taped or patched, then re-rolled. Sealant is a cap on a mechanical termination, not a fix for a failed seam, and a roof patched with caulk over bad welds is a roof that fails on schedule.

Then the final. After the repairs, the marked spots get re-inspected, the punch list gets closed item by item, and the seam log gets completed before the membrane work is signed off. On a warranted system, this internal walk happens before the manufacturer's rep ever shows up, because finding and fixing your own thin spots is far cheaper than having the rep find them and hold the warranty. The crew that probes its own roof keeps the warranty. The crew that waits for the rep gambles with it.

The warranty inspection and the NDL warranty

On a manufacturer-warranted single-ply roof, the manufacturer's field inspection is what releases the warranty, and the seams are the heart of it. A no-dollar-limit warranty, the NDL, is the strong one: it covers the full cost of repair on the covered roof system for the term, commonly 10 to 30 years with 20 years typical, with no cap that erodes over time. The manufacturer does not hand that out on trust. A technical rep inspects the completed roof against the published system before the warranty issues.

The inspection is built around the details that decide whether the roof lasts. The rep confirms the installer is an authorized, trained applicator using the manufacturer's products throughout, then checks the seams, the flashings, the terminations, and the penetrations. Expect destructive seam samples cut and peeled for film-tearing bond, expect the probe, and expect the corners and T-joints to get the hardest look. A roof that reads light on a seam sample or shows a rushed corner fails the check no matter how good the field looks.

Two more realities ride with the NDL. The warranty almost always requires every product in the assembly to be the warranting manufacturer's, installed to their specification, which is why mixing in a cheaper boot or sealant can void coverage. And most NDL warranties require documented periodic maintenance inspections over the life of the roof, at the owner's expense, or the coverage lapses. Read the actual warranty language on the system you are installing, because the conditions, not the headline term, are what get tested when a claim comes in.

What to document

The seam record is what defends the roof when a leak shows up and the question becomes who pays. A warranty claim turns on whether the seams were welded right, tested, and repaired, and the only proof is the log made while the work happened. The seams cure or set the day the crew leaves, so the daily trial weld, the probe findings, and the repairs are the evidence the roof was built to spec.

Capture it by area as the work goes: the roof section, the membrane type and thickness, the finished weld width measured, the daily trial-weld result and whether it pulled to film-tearing bond, the probe findings and where, the destructive sample results, every repair and where it was, the flashing and detail checks, and the conditions, temperature, wind, and surface state at welding. Tie the whole package to the warranty type and the manufacturer's inspection, because that is the file the rep reads and the file a claim is judged against years out.

Common mistakes

• Skipping the daily trial weld, so the welder runs all day on settings nobody confirmed against the conditions.
• Running welds narrower than the manufacturer's minimum, commonly 1.5 in, with no cover-strip overlay.
• Welding too cold or too fast, leaving cold welds that probe open a season later.
• Welding over a dirty, chalky, or damp membrane, so a boundary layer stops the sheets from fusing.
• Leaving T-joints untreated, so the three-layer step at every seam crossing channels water.
• Installing EPDM splice tape with no primer or no roller, so the tape never fully bonds to the cured rubber.
• Probing only a sample of the seams, or probing warm seams and tearing good welds.
• Over-welding until the membrane smokes and bleeds, degrading the polymer the weld was supposed to fuse.
• Rushing the flashings, pipe boots, and field-fabricated corners at the end of the shift, where most leaks start.
• Calling the roof done before the manufacturer's seam samples, probe, and detail check confirm the warranty.

Field checklist

Standards and references

The document that governs the weld, the seam, and the warranty is the membrane manufacturer's published specification and warranty, full stop. It sets the weld temperature and speed range, the finished weld-width minimum, the trial-weld and probe requirements, the flashing and T-joint details, and the conditions that decide whether the warranty issues. The closeout inspection is run against it, so confirm every figure in this guide against the actual product literature.

The ASTM material specifications define what each membrane has to be. ASTM D6878 covers thermoplastic polyolefin (TPO) sheet roofing, ASTM D4434 covers polyvinyl chloride (PVC) sheet roofing, and ASTM D4637 covers EPDM sheet for single-ply membranes. These standards specify the sheet, and separate ASTM test methods cover seam strength and the peel and shear tests behind the destructive sample. Confirm the current designation and edition before citing one on a submittal, because these get revised across cycles.

Around the material standards sits the rest of the framework. The NRCA Roofing Manual is the practical reference for single-ply installation and seam practice. Wind uplift runs under ASCE 7 for the loads, with FM Global data sheets adding their own requirements on insured roofs and SPRI publishing wind-design standards for single-ply systems, including the ballasted standard. The building code adopts these by jurisdiction and amends them, so confirm the requirement against the adopted edition and the project specification, and let the manufacturer's warranty override any rule of thumb on the weld and the seam.

Units, terms, and conversions

Single-ply seam work uses a small, specific vocabulary, and the same idea reads differently across a spec, a data sheet, and a warranty, so the terms are worth pinning down.

Membrane thickness is given in mils, thousandths of an inch, with common single-ply at 45, 60, or 80 mil. Weld width and squeeze-out are in inches, with the finished weld commonly at least 1.5 in. Welder temperature is in degrees F at the nozzle and welder speed in feet per minute. A hot-air weld is a fusion bond on a thermoplastic; a splice is an adhesive bond on EPDM. Film-tearing bond, FTB, is the acceptance where the sheet fails before the seam. A T-joint is where two seams cross. An NDL warranty is the no-dollar-limit, full-cost manufacturer warranty.

TPO / PVC

Thermoplastic single-ply membranes whose seams are hot-air welded into one fused material

EPDM

Thermoset rubber single-ply membrane whose seams are made with splice tape and primer, not welded

Hot-air weld

A seam fused by hot air and roller pressure, melting the two thermoplastic sheets into one

Film-tearing bond (FTB)

The weld acceptance where a peeled sample tears the membrane before the seam opens

Seam probe

A non-destructive test drawing a blunt tool along a cooled weld to find voids and cold welds

Trial weld

A daily and per-condition test seam pulled by hand to confirm welder settings before production

T-joint

The three-layer point where two seams cross, the highest-risk spot on a field seam

NDL warranty

No-dollar-limit warranty covering full repair cost on the system, issued after the manufacturer's inspection

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What is a roof cricket?
• What is the difference between wet mils and dry mils?
• What is a ballasted roof?
• What is an EPDM roof?
• TPO vs EPDM vs PVC: which single-ply membrane is best?